Home

Search

Collections

Journals

About

Contact us

My IOPscience

Magnetic and structural relationship of RFe2Si2 and R(Fe1−x Mx )2Si2(x = 0−1) systems (R = La, Y and Lu, M = Ni, Mn and Cu)

This content has been downloaded from IOPscience. Please scroll down to see the full text. 2014 J. Phys.: Condens. Matter 26 476002 (http://iopscience.iop.org/0953-8984/26/47/476002) View the table of contents for this issue, or go to the journal homepage for more

Download details: IP Address: 128.187.103.98 This content was downloaded on 18/05/2015 at 13:43

Please note that terms and conditions apply.

Journal of Physics: Condensed Matter J. Phys.: Condens. Matter 26 (2014) 476002 (11pp)

doi:10.1088/0953-8984/26/47/476002

Magnetic and structural relationship of RFe2Si2 and R(Fe1−xMx)2Si2(x = 0 − 1) systems (R = La, Y and Lu, M = Ni, Mn and Cu) I Felner1 , Bing Lv2 and C W Chu2 1

Racah Institute of Physics, The Hebrew University of Jerusalem, Jerusalem, 91904, Israel Department of Physics and the Texas Center for Superconductivity, University of Houston, Houston, TX 77204-5005, USA

2

E-mail: [email protected] Received 19 June 2014, revised 17 September 2014 Accepted for publication 30 September 2014 Published 27 October 2014 Abstract

Due to the similarity between AFe2 As2 (A = Ba, Sr) and RFe2 Si2 (R = La, Y and Lu), the RFe2 Si2 system has been proposed as a potential candidate for a new high TC superconducting family containing Fe–Si (instead of Fe–As) layers as a structural unit. Various R (Fe1−x Mx )2 Si2 (M = Ni and Cu) materials were synthesized and measured for their magnetic properties. None of these materials is superconducting down to 1.8 K. A pronounced peak at 232 K was observed in the magnetization curve of YFe2 Si2 . 57 Fe M¨ossbauer studies confirm the absence of any long-range magnetic ordering below 232 K. Similar peaks at various temperatures also appear in R (Fe1−x Mx )2 Si2 samples. For Y (Fe1−x Mnx )2 Si2 the peak position is dramatically affected by the magnetic Mn dopants. Four independent factors affect the peak position and shift it to lower temperatures: (i) the lattice parameters, (ii) the concentration of x (iii) the applied magnetic field and (iv) the magnetic nature of M. It is proposed that the magnetic peaks observed in RFe2 Si2 and in R (Fe1−x Mx )2 Si2 represent a new nearly ferromagnetic Fermi liquid system, its nature is yet to be determined. Keywords: magnetic properties, rare-earth silicides, nearly ferromagnetic Fermi liquid, superconductivity (Some figures may appear in colour only in the online journal)

The recent discovery of superconductivity (SC) at relative high temperatures in F doped SmFeAsO (1 1 1 1), up to TC = 55 K [4], as well as in the doped AFe2 As2 (A = Ba, Sr) (1 2 2) systems, has stimulated a large number of experimental and theoretical studies and intensified the search for high temperature superconductors (HTSC) in materials containing Fe–As layers as a structural unit. The pristine bct BaFe2 As2 sample, has a spin-density-wave (SDW) ground state at TN = 136(1) K. The suppression of the SDW state by doping, in most cases, induces SC in the system. Notably, SC is generated by both electron and hole doping and even by iso-valent atomic substitution, e.g. P for As. Subsequently, partial substitution of Ni or Co for Fe in BaFe2 As2 , induces

1. Introduction

Over the last four decades the ternary intermetallic compounds, which crystallize in the body-centered tetragonal (bct) ThCr2 Si2 (space group I 4/mmm), have been of great interest due to the variety of physical phenomena observed in these compounds. As early as 1974, both magnetic and 57 Fe M¨ossbauer effect spectroscopy (MS) studies suggest that in bct RFe2 M2 (R = La, Y and Lu and any other rare-earth element, M = Si or Ge), the Fe ions are not magnetically ordered [1]. Indeed, neutron powder diffraction measurements on NdFe2 Si2 confirmed the absence of any magnetic moment on the Fe sites [2, 3]. 0953-8984/14/476002+11$33.00

1

© 2014 IOP Publishing Ltd Printed in the UK

J. Phys.: Condens. Matter 26 (2014) 476002

I Felner et al

The density of states (DOS) at the Fermi level strongly depend on the distortion of the FeX4 (X = Si and/or Ge) tetrahedra and/or the height of the Si (Ge) atom from the two-dimensional Fe plane. A steep increase of the DOS of Fe is observed below the Fermi energy. It is claimed that this condition, creates excellent chances that doping in the Fe site will shift the DOS to the Fermi level, condition which is essential for HTSC [14]. The desire to dramatically change or tune the properties of any compound, preferably through small changes in stoichiometry or composition, motivated us to search for new HTSC in doped RFe2 Si2 (R = La, Y and Lu) systems in which R is a non-magnetic rare-earth element. Similar to SC induced in doped BaFe2 As2 system, it was expected that partial substitutions of R, Fe or Si in RFe2 Si2 will induce SC at elevated temperatures. Therefore, substitution was made in all three elements which all crystallize in the tetragonal ThCr2 Si2 type structure. Here, we present only the magnetic properties of R (Fe1−x Mx )2 Si2 , in which doping was done in the Fe site. All other systems such as: Ce4+ substituted for La3+ , or magnetic Ho substituted for Y, will be described elsewhere. For all R = (La, Y and Lu), Ni was substituted for Fe, whereas in YFe2 Si2 the Fe atoms were also partially replaced by Cu and Mn. In this tetragonal structure both Ni and Cu atoms are nonmagnetic [15] and our magnetic measurements show that none of these compounds exhibits SC traces down to 1.8 K, although SC in Fe or Ni-containing 122 compounds was observed at low temperatures, e.g. YFe2−x Si2 (TC = 3.0 K) [16], BaNi2 P2 (TC = 3.0 K) [17] and LaNi2 P2 (TC = 1.8 K) [18]. Unexpectedly, pronounced magnetic peaks appear at various temperatures, which are affected by the applied magnetic field (H ) and/or by the dopant concentrations. For example: (i) the peak observed in YFe2 Si2 at 232 K (measured at H = 15 Oe), is shifted to 220 K when measured at H = 1 kOe. (ii) In Y (Fe1−x Nix )2 Si2 , the peak shifts with x to lower temperatures and for YNi2 Si2 (x = 1) no noticeable peak is observed. Its magnetic behavior is reported here for the first time. On the other hand, in Y (Fe1−x Cux )2 Si2 , the peak at T > 300 K (for x = 0.1), is shifted back to lower temperatures for higher x values. Since both R and Si are non-magnetic elements, the magnetic peaks are definitely related to the Fe sites. Our 57 Fe M¨ossbauer spectroscopy (MS) studies on various materials indicate clearly the absence of any permanent long-range ordering of Fe, below the peaks position. We propose that the peaks observed in RFe2 Si2 and in their derivatives represent a new nearly ferromagnetic (FM) Fermi liquid (NFFL) systems, their nature will be discussed.

Figure 1. The body-centered-tetragonal crystal structure of the

RFe2 Si2 .

SC in the Ba (Fe1−x Nix )2 As2 and the Ba (Fe1−x Cox )2 As2 systems [5–12]. A higher value for TC = 38 K, was observed by optimal doping in the Ba1−x Kx Fe2 As2 system [5] and in pristine BaFe2 As2 by application of high pressure [9]. Also associated with or preceding the magnetic transition is a tetragonal to orthorhombic structural transition, which is suppressed in the SC state. Perhaps, even more remarkable than the large TC , is the large tunability these systems possess. It should be noted, that in the layered AFe2 As2 systems (for non-magnetic A atoms such as Ba or Sr) suppression of the long-range magnetic ordering of the Fe sublattice is necessary for the appearance of SC, similar to the behavior of the HTSC cuprates. On the other hand, the problem of these systems is the toxicity of As which limited their fabrication. The above discoveries motivated a search for new superconductors at elevated temperatures of other systems having the same ThCr2 Si2 -type structure. In the tetragonal RFe2 Si2 (as well as in BaFe2 As2 at elevated temperatures) (figure 1) the R(Ba), Fe and Si(As) ions reside in the 2a, 4d and 4e crystallographic positions, respectively. The close relation between the Fe–As and Fe–Si(Ge) systems, strongly suggests similar major role of Fe in the occurrence of SC. Therefore an extensive search for HTSC at elevated temperatures in the two similar RFe2 Si2 and RFe2 Ge2 systems is appealing. In both systems the Fe ions are not magnetic, thus no suppression of the SDW state mentioned above is needed. In addition both Si and Ge are cheap, non-toxic and very convenient materials. Indeed, very recently YFe2 Ge2 was found to be SC at TC ∼ 1.8 K in [13]. That strengthens our intuition and prediction for extensive searching for new superconductors in these two systems. Up to date, little information is available on the electronic structure of RFe2 Si2 and RFe2 Ge2 . The electronic structures of LaFe2 Si2 and LaFe2 Ge2 were calculated from first principles [14]. Despite of the almost two-dimensionality of the crystal structure, the Fermi surface is three-dimensional.

2. Experimental details

Polycrystalline samples with nominal composition La(Fe1−x Nix )2 Si2 (x = 0, 0.05 and 0.15), Y (Fe1−x Nix )2 Si2 (x = 0, 0.05, 0.15, 0.50, 0.75 and 1), Y (Fe1−x Mnx )2 Si2 (x = 0.10, 0.20, 0.30, 0.5, 0.7 and 1), Y (Fe1−x Cux )2 Si2 (x = 0.10, 0.25 and 0.5) and Lu (Fe1−x Nix )2 Si2 (x = 0, 0.1, 0.2 and 0.3), were prepared by melting stoichiometric amounts of R, Fe and Si with Ni, Mn or Cu (all of at least 99.9% purity) in an arc furnace under high-purity Ar atmosphere. The arc-melted buttons were 2

J. Phys.: Condens. Matter 26 (2014) 476002

I Felner et al

A typical XRD pattern obtained for YFe2 Si2 is shown in (figure 2). Rietveld refinement on a long term XRD run yields z = 0.371(1) for the free parameter of Si in the 4(e) site. The lattice parameters for YFe2 Si2 are in fair agreement with a = 3.910 and c = 9.92 Å published in [19]. EDS chemical analysis of YFe2 Si2 shows that the Y : Fe : Si ratio is exactly 1 : 2 : 2. The R (Fe1−x Nix )2 Si2 system was extensively investigated and the XRD patterns indicate single-phase materials which crystallize in the tetragonal ThCr2 Si2 structure. For small x values the a lattice is practically constant, whereas the regular decrease in the c lattice constant can be attributed to the smaller atomic radii of Ni (1.49 Å) as compared to that of Fe (1.56 Å) [20]. That leads to the decrease of unit cell volume with x (table 1) and to the decrease of c/a ratio in Y (Fe1−x Nix )2 Si2 as shown in figure 2 (inset). EDS studies of Y (Fe1−x Nix )2 Si2 confirm the nominal compositions of the Ni-doped samples; e.g. for x = 0.05, and 0.15 the measured Ni concentrations are: 0.06 and 0.17 respectively. On the other hand, table 1 shows that the unit cell volume for Y (Fe1−x Mnx )2 Si2 and Y (Fe1−x Cux )2 Si2 increases with x. For Mn (atomic radius 1.61 Å) this increase is quite obvious, whereas for Cu (atomic radius 1.45 Å) [20] this increase needs more consideration and will be discussed later. In YFeCuSi2 (x = 0.5), additional extra lines in the XRD pattern which account to 5–7% of the spectral area, were observed. These lines can be indexed on a basis of a hexagonal structure (AlB2 type) which belong to the nonmagnetic YCuSi with a = 4.031 and c = 4.008 Å [21]. Extra lines which account to 15% of the spectral area were observed in YFeMnSi2 (x = 0.5). They belong to an orthorhombic unit cell (SG Cmcm) with a = 4.047, b = 15.83 and c = 3.924 Å, similar to the structure of YFe0.33 Si2 .

Figure 2. XRD pattern of YFe2 Si2 . The inset shows the c/a ratio for the Y (Fe1−x Nix )2 Si2 system.

flipped and re-melted several times to ensure homogeneity. The samples were structurally and chemically characterized by a Panalytical X’pert powder x-ray diffraction (XRD) diffractometer and energy dispersive spectroscopy (EDS) using EDS-JOEL JSM-7700 scanning electron microscopy (SEM). The EDS instrumental error is ∼2–3%) The XRD patterns of all samples could be well indexed on the basis of bct I 4/mmm type structure. Magnetization measurements at various applied magnetic fields in the temperature interval 1.8 K < T < 350 K have been performed using the commercial (Quantum Design) superconducting quantum interference device (SQUID) magnetometer, with samples mounted in gel-caps. Prior to recording the zero-field-cooled (ZFC) curves, the SQUID magnetometer was always adjusted to be in a ‘true’ H = 0 state. The temperature dependence of the field-cooled (FC) and the ZFC branches were taken via warming the samples. The sharp peaks were assigned as the temperatures in which the maximum magnetic moments were observed. For the broad peaks, we used the full-width at halfmaximum (FWHM) method. The real (χ  ) and imaginary (χ  ) ac susceptibilities were measured with a home-made pickup coil method at a field amplitude of h0 = 0.05 Oe at frequencies of ω/2π = 56 and 1465 Hz. Resistivity measurements down to 1.8 K have been performed by conventional four probe methods. 57 Fe MS studies at 90 and 295 K (RT) were performed using a conventional constant acceleration drive, with a 50 mCi 57 Co : Rh source. The spectra were analyzed in terms of least-squares fit procedures to theoretical expected spectra. The velocity calibration was performed with α-Fe foil at RT and the reported isomer shift (IS) values are relative to this foil at RT.

3.2. Magnetic studies of RFe2 Si2 (R = Y, La and Lu)

Comprehensive magnetic measurements have been performed on the parent RFe2 Si2 materials and on all Ni and Cu doped samples listed above. No traces for SC were detected down to 1.8 K. A typical resistivity study (at ambient pressure) for YFe2 Si2 also indicates the absence of SC down to 1.8 K (figure 3, lower panel). On the other hand, the parent YFe2 Si2 and LuFe2 Si2 compounds exhibit pronounce magnetic peaks at various temperatures in the magnetization M(T ) curves. Generally speaking, the magnetic features of the Ni and Cu doped materials are very similar to that of their parent compounds. For the sake of clarity, we start with the data measured for YFe2 Si2 . Figure 3 shows the ZFC magnetization curve (M(T )) of YFe2 Si2 measured at H = 250 Oe, in which the peak at 226 K is readily observed. The peak position (which is defined as the maximum of M) depends strongly on H . Figure 3 (inset) shows (a) that for H = 1 kOe the peak shifts to 220 K and (b) that the same position is obtained regardless of whether the M(T ) was measured via ZFC or FC processes. The bifurcation observed at low temperatures is probably due to a tiny FM phase discussed hereafter. The field dependence (i) YFe2 Si2 .

3. Experimental results 3.1. Structural and chemical characterization of R(Fe1−x Nix )2 Si2 and Y(Fe1−x Cux )2 Si2 samples

All XRD patterns obtained were indexed on the basis of a bct structure with the lattice constants given in table 1. 3

J. Phys.: Condens. Matter 26 (2014) 476002

I Felner et al

Table 1. Lattice parameters, unit cell volume, magnetic peak position at 250 Oe and the spontaneous magnetization (MS ) at 5 K of R (Fe1−x Mx )2 Si2 .

Compound

a(Å) ±0.001

Y (F e1−x Nix )2 Si2 x=0 3.923 x = 0.05 3.920 x = 0.15 3.923 x = 0.50 3.930 x = 0.75 3.943 x = 1.0 3.963 Y (F e1−x Cux )2 Si2 x = 0.10 3.924 x = 0.50c 3.949 Y (F e1−x Mnx )2 Si2 x = 0.10 3.923 x = 0.20 3.925 x = 0.30 3.921 x = 0.50c 3.921 Lu (F e1−x Nix )2 Si2 x=0 3.875 x = 0.10 3.876 x = 0.20 3.878 x = 0.30 3.881 La (F e1−x Nix )2 Si2 x=0 4.059 x = 0.05 4.054 x = 0.15 4.054

c(Å) ±0.003

V (Å)3

Peak at 250 Oe ± 2 K

MS (5 K) (emu g−1 )a

9.951 9.912 9.874 9.76(1) 9.73(3) 9.544

153.1 152.3 151.9 150.7 151.2 149.8

226 211 192 54b 49b —

0.31(2) 0.42 0.13 1.82 2.21

9.94(1) 9.90(2)

153.0 154.4

>350 170

0.008 0.014

9.963 9.759 10.013 10.129

153.3 153.6 153.9 155.7

61 16 —

9.869 9.824 9.790 9.761

148.2 147.6 147.2 147.0

70 46 31 23

0.001 — 0.002 0.013

10.163 10.130 10.10

167.4 166.5 166.0

>350 >350 >385

0.96(1) 1.63 2.34

a

MS are remnant values due to FM impurity phases. Determined by FWHM. c Contain extra lines. b

of the peak position for YFe2 Si2, is shown in figure 4. The peak obtained at 232 K for H = 15 Oe, shifts to 118 K for H = 10 kOe (main panel) and further to 209 K, 197 K and 186 K for H = 20 kOe, 30 kOe and 45 kOe (inset) respectively. The almost linear field dependent of the peak position yields a slope of 1.2 K kOe−1 . It should be noted, that no observable peak was obtained by ac susceptibility studies. In addition, figure 3 (lower panel) demonstrates that no distinct peak is visible in ρ(T ) plot measured up to 300 K. However, a slope change around 230–240 K is observed (figure 3). The peak origin is discussed later. In order to clarify the nature of these peaks, 57 Fe MS below and above (at 90 and 295 K) the peak position have been carried out. Figure 5 shows two almost identical narrow singlet obtained with a line width of 0.333(3) mm s−1 . Due to the typical second order Doppler shift the two spectra differ only in their IS values, which are 0.182 ± 0.002 mm s−1 and 0.195 ± 0.002 mm s−1 for RT at 90 K respectively. Although the Fe ions in the 4d crystallographic site have a non-cubic symmetry, the quadrupole splitting, is relatively small, 0.06±0.02 mm s−1 . These values correspond to divalent non-magnetic Fe state found in other Fe-based intermetallic compounds with the ThCr2 Si2 -type structure [22, 23]. Figure 5 definitely proves the absence of sizable permanent long-range magnetic moments in the Fe sites, and confirms our old claim that Fe in RFe2 Si2 is non-magnetic [1–3].

The isothermal magnetization M(H ) curves of YFe2 Si2 measured at 5 and 295 K are shown in figure 6. These curves can be fitted as M (H ) = MS + χp H , where MS is the spontaneous FM magnetization and χp H is the linear paramagnetic (PM) intrinsic susceptibility. For both temperatures, MS = 0.31(2) emu g−1 (table 1) is attributed to a tiny FM extra phase. This clearly indicates that for the FM impurity the magnetic temperature ordering (TC ) is well above RT and that its presence does not affect the peak positions. This impurity probably corresponds to ∼0.14% unreacted pure Fe (MS = 220 emu g−1 , TC = 1040 K), or alternatively, to the binaries: Fex Si (Fe3 Si, TC = 808 K) [24] and/or YFe2 or YFe3 (TC = 542 and 525 K) [25]. The tiny amount of this FM impurity phase is below the detection limit of XRD and MS techniques. The hysteresis loop at 5 K (shown in figure 6 (inset)) with a coercive field (HC ) of 245 Oe reflects this FM phase. The M(T ) curve at 250 Oe (figure 3), is the net moment values after subtracting the FM contribution. Well above the peak position the M(T ) curves exhibit a typical PM shape and adhere closely to the Curie–Weiss (CW) law: χ (T ) = χ0 + C/(T − θ), where χ (= M/H ), χ0 is the temperature-independent part C is the Curie constant, and θ is the CW temperature. The PM parameters extracted from M(T ) measured at 1 kOe (at 260–350 K) are: χ0 = 5.5 × 10−3 emu mol−1 Oe−1 , C = 2.34(1) emu K mol−1 Oe−1 and θ = 191(3) K. The χ0 infers to the temperature independence FM impurity as discussed. This C value corresponds to a PM 4

J. Phys.: Condens. Matter 26 (2014) 476002

20

50

M (emu/ mol)

Moment (emu/mol)

25

I Felner et al

1 kOe 40

30

226 K

220 K

FC ZFC

15

100

200

300

Temperature (K)

YFe 2Si2

10

250 Oe

0

100

200

300

Temperature (K) 57

Fe M¨ossbauer studies of YFe2 Si2 at 90 K (main panel) and at 295 K (inset). Figure 5.

Figure 3. ZFC magnetization curve of YFe2 Si2 (after subtraction of the FM contribution) measured at 250 Oe. The inset shows the ZFC and FC branches measured at 1 kOe. The lower panel exhibits the low temperature (down to 1.8 K) resistivity measurements at ambient pressure.

Figure 6. Isothermal magnetization M(H ) plots measured at 5 and 295 K for YFe2 Si2 . The upper inset is the hysteresis loop measured at 5 K. The lower inset shows the two components at 5 K deduced from the same M(H ) plot.

to virtually temperature-independent magnetic susceptibility (Pauli PM) claimed in the past [1]. Note, that the measured peak size magnitude for YFe2 Si2 at 250 Oe is ∼10−2 emu, which is below the sensitivity of the old magnetometer used in [1]. The ZFC magnetic behavior of LaFe2 Si2 is a bit different. In contrast to YFe2 Si2 (figure 3), the magnetization measured at H = 250 Oe increases up to 350 K but no definite peak is observed (figure 7). Presumably, higher temperatures are needed to reveal this peak as observed for the iso-structural LaNi2 Ge2 material [27]. A small bump around 225 K is observed, its nature is yet not known. On the other hand, for LuFe2 Si2 the peak at 70 K under H = 250 Oe is clearly observed. The isothermal magnetization curves at 5 K of LaFe2 Si2 and LuFe2 Si2 (figure 7 lower inset) can be fitted in the same manner as the M(H ) plots of YFe2 Si2 (figure 6). The deduced FM MS values are: 0.96 and 0.001 emu g−1 respectively (table 1). The almost straight line at 5 K for LuFe2 Si2 excludes the existence of the SC Lu3 Fe3 Si5 phase (ii) LaFe2 Si2 and LuFe2 Si2 .

Figure 4. The field dependence of the peak position of YFe2 Si2 . The inset shows this behavior for high applied fields.

effective moment of Peff = 3.06(2)µB /Fe, a value which is very similar to 2.9µB /Fe obtained for LuFe2 Ge2 [26]. χ (T ) obtained fits well with χp deduced at 5 K. It appears that the Fe in YFe2 Si2 carries a net PM moment. That is in stark contrast 5

J. Phys.: Condens. Matter 26 (2014) 476002

I Felner et al

Figure 7. ZFC magnetization curve of LaFe2 Si2 and LuFe2 Si2

Figure 9. M(T ) plot of Y(Fe0.5 Ni0.5 )2 Si2 measured at 250 Oe. The inset shows the M(T ) curve measured at 20 Oe.

(upper inset) measured at 250 Oe. The lower inset shows the isothermal M(H ) plots of LaFe2 Si2 and LuFe2 Si2 measured at 5 K. Note the almost linear curve of LuFe2 Si2 .

Figure 10. Temperature dependence of the magnetization of Y(Fe0.25 Ni0.75 )2 Si2 measured at 250 Oe. The inset shows the peak position of all Y (Fe1−x Nix )2 Si2 samples measured at 20–30 Oe. Figure 8. ZFC magnetization curve of Y (Fe1−x Nix )2 Si2 (x = 0, 0.05 0.15 (inset)) measured at 250 Oe.

contain an extra FM phase similar to that shown in figure 6. The differences among the samples are (i) the peak position which is strongly depends on x, (ii) the spread MS values as listed in table 1. Y (F e1−x N ix )2 Si2 . (i) Figures 8–10 show representative M(T ) plots of Y (Fe1−x Nix )2 Si2 samples. Figure 8 shows the M(T ) curve for x = 0, 0.05 and 0.15 all measured at 250 Oe. The peak at 226 K for YFe2 Si2 (x = 0) shifts to 211 and to 193 K (inset) for x = 0.05 and 0.15 respectively. Broad peaks (determined by the FWHM methods) are observed for higher Ni concentrations: x = 0.5 and 0.75 at 54 and 49 K respectively (table 1). (ii) For each Y (Fe1−x Nix )2 Si2 material, the peak is also field depended. For example: (a) for x = 0.05, the peak at 220 K under 15 Oe shifts to 211 K (figure 8) and to 193 K when measured at 250 and 1 kOe respectively. (b) For x = 0.15, the peak at 193 K, measured at 250 Oe (figure 8 inset), shifts to 180 K when measured at 1 kOe. (c) For x = 0.5 the relatively broad peak at 86 K measured at 20 Oe (figure 9 inset), shifts at 250 Oe to a broad peak its maximum is at 54 K (table 1). The peak position of Y (Fe1−x Nix )2 Si2 samples measured at 15–25 Oe is shown in figure 10 (inset). (iii) All M(H ) plots measured below and above the peak position are

(TC = 6 K) [28]. The main goal in presenting the M(T ) curves for the three RFe2 Si2 samples (figures 3 and 7), is to show the appearance of pronounced magnetic peaks and their locations. Therefore, in figure 7, no subtraction of the impurity FM phase contribution was made. Since in all RFe2 Si2 (R = La, Y and Lu) R are not magnetic elements, the only difference between the three samples is in their unit-cell dimensions. Table 1 shows that the lattice parameters as well as the unit cell volumes decrease as: LaFe2 Si2 > YFe2 Si2 > LuFe2 Si2 . The same trend is observed in peak position (measured at the same H ), indicating clearly the connection between the lattice dimensions and the peak positions. 3.3. Magnetic studies of R (Fe1−x Nix )2 Si2

The magnetic behaviors of all R (Fe1−x Nix )2 Si2 samples studied resemble the features observed in their parent compounds and the relevant data obtained are listed in table 1. For each x value, the M(T ) curves show distinct peaks at elevated temperatures which decrease with H . All M(H ) plots 6

J. Phys.: Condens. Matter 26 (2014) 476002

I Felner et al

Figure 11. Temperature dependence of the magnetization of YNi2 Si2 measured at 250 Oe. The inset shows the almost linear isothermal magnetization at 5 K.

Figure 12. Temperature dependence of the magnetization of La(Fe0.85 Ni0.15 )2 Si2 measured at 10 Oe. The inset shows the 57 Fe M¨ossbauer spectrum at 90 K.

similar to these obtained for YFe2 Si2 (figure 6) and can be fitted by the same manner. The spread MS (at 5 K) values deduced are listed in table 1. Note that the highest MS = 2.21 emu g−1 obtained (for x = 0.75), is equivalent to the saturation moment of ∼1% unreacted Fe, a level which is undetectable neither by XRD nor by MS. An additional point of interest is the magnetic behavior of YNi2 Si2 (figure 11) which is presented here for the first time. The lattice parameters obtained for YNi2 Si2 (table 1) are in fair agreement with data published in [19]. The M(T ) curve depends on the temperature (not a Pauli behavior), but all attempts to fit it to the CW law were unsuccessful. Both M(H ) curve measured at 5 K and at 250 K are linear and the calculated slope χ (= M/H ) value at 5 K and 246 K are 3.76×10−4 emu mol−1 Oe−1 and 1.96×10−4 emu mol−1 Oe−1 respectively. This linear behavior strengthens our suggestion that the undetermined extra FM phase is due to pure Fe or to any other Fe intermetallic compound. La (F e1−x Nix )2 Si2 and Lu (F e1−x N ix )2 Si2 . In these two systems only materials with low Ni concentration have been synthesized. Their M(T ) and M(H ) plots are very similar to their parent compounds and the relevant data obtained are listed in table 1. Although the peak position strongly depends on x, no peak is observed in La(Fe0.85 Ni0.15 )2 Si2 up to 385 K. Here again a bump in the M(T ) is observed around 260 K (figure 12). Table 1 shows that for La (Fe1−x Nix )2 Si2 the deduced MS values (at 5 K) increases with x. The57 Fe MS measured at 90 K (figure 12 inset) exhibits a narrow singlet with almost the same parameters obtained for YFe2 Si2 . This proves once again the absence of sizable permanent local magnetic moments in the Fe sites. In Lu (Fe1−x Nix )2 Si2 (similar to Y (Fe1−x Nix )2 Si2 ) the peak position shifts with x to lower temperatures (figure 13). At 250 Oe, the peaks for x = 0.1 and 0.2 are at 46, 31 K respectively (table 1). For x = 0.2, the peak is shifted to 43 K when measured at 25 Oe. For both materials, an additional broad peak is observed at 205 K (not shown) its origin is yet to be determined. For x = 0.3, at 100 Oe

Figure 13. Temperature dependence of the magnetization of Lu(Fe0.7 Ni0.3 )2 Si2 and Lu(Fe0.9 Ni0.1 )2 Si2 (inset) measured at 100 Oe and 250 Oe respectively.

the peak location is at 31 K (figure 13) and it is shifted to 18 K when measured at 1 kOe. At higher temperatures the M(T ) curve for x = 0.3 also exhibit a typical PM shape and can be fitted by the CW law. The extracted PM parameters (at 1 kOe) are: χ0 = −4.2 × 10−3 emu mol−1 Oe−1 , C = 2.60(2) emu K mol−1 Oe−1 and θ = −70(1) K. This C value corresponds to Peff = 3.71µB /Fe and proves once more the PM nature of Fe in the RFe2 Si2 systems. The M(H ) curves of Lu(Fe1−x Nix )2 Si2 are almost linear and the deduced MS values are small and negligible (table 1). 3.4. Magnetic studies of Y (Fe1−x Cux )2 Si2

The change in lattice parameters Y (Fe1−x Cux )2 Si2 is a bit different than that of R (Fe1−x Nix )2 Si2 systems. Table 1 shows that for Cu the a lattice parameter increases with x whereas the c constant practically remains unchanged. This 7

J. Phys.: Condens. Matter 26 (2014) 476002

I Felner et al

Figure 14. Temperature dependence of the magnetization of Y(Fe0.9 Cu0.1 )2 Si2 measured at 250 Oe. The inset shows the linear isothermal magnetization at 5 K.

Figure 16. Temperature dependence of the magnetization of Y(Fe0.8 Mn0.2 )2 Si2 and Y(Fe0.9 Mn0.1 )2 Si2 (inset) both measured at 250 Oe.

Thus, the applied magnetic field affects the Y (Fe1−x Cux )2 Si2 system in the same manner. All M(H ) curves are almost linear (figures 14 and 15 (insets)) and the FM impurities are negligible. This indicates once again that the peaks observed are intrinsic and not from FM impurity phases. The extra hexagonal YCuSi phase in YFeCuSi2 (detected by XRD) is not magnetic and cannot account for peaks observed in figure 15 [29]. Magnetic studies of Y (Fe1−x Mnx )2 Si2

The RMn2 Si2 system was extensively studied in the past. All compounds are magnetically ordered at high temperatures and the interactions within the layers between the Mn–Mn moments are always FM. The nature of the interlayer exchange interaction is FM or AFM, depending on the lattice constants size. e.g. LaMn2 Si2 is FM with TC = 310 K and becomes AFM ordered up to 470 K whereas YMn2 Si2 is AFM with TN = 510 K [30, 31]. Using the 57 Fe MS we have shown that (i) in the doped materials Fe has no magnetic moment of its own and reveals the magnetic order of the Mn sublattice through transferred hyperfine field. (ii) In contrast to the systems described above, Mn in Y (Fe1−x Mnx )2 Si2 carries a local magnetic moment [31]. The major goal here is study the effect of magnetic Mn on peak positions in YFe2 Si2 . XRD studies on Y (Fe1−x Mnx )2 Si2 (x = 0.1, 0.2, 0.3,0.5, 0.7 and 1) show that single-phase materials could be obtained for low Mn compositions only. In YFeMnSi2 (x = 0.5) a few extra lines which belong to the orthorhombic YFe0.33 Si2 type structure have been observed. In addition, for x = 0.7 and YMn2 Si2 (x = 1) extra lines which belong to the AFM Y2 Mn3 Si5 (TN = 96 K) phase were detected [32]. Therefore, their magnetic properties will not be discussed here. The effect of Mn substitution on the peak position is more dramatic. Figure 16 (inset) exhibits the M(T ) curve for Y(Fe0.9 Mn0.1 )2 Si2 measured at 250 Oe in which the peak at 63 K is well observed. Although, the lattice parameters of this material and pure YFe2 Si2 are very similar (table 1), the

Figure 15. Temperature dependence of the magnetization of Y(Fe0.5 Cu0.5 )2 Si2 measured at 30 Oe. The inset shows the linear isothermal magnetizations at 5 and 250 K which are below and above the peak position.

agrees well with the higher a lattice constant of YCu2 Si2 (a = 4.153 and c = 9.92 Å) as compared to that of YFe2 Si2 [19]. In the ThCr2 Si2 type structure the shortest 3d–3d distance is given by given by 2−1/2 a, which means that the Cu size is larger than that of Fe. That is due to the well accepted determination that in many intermetallic compounds Cu is formally monovalent, in contrast to a formal divalent Fe (and also Ni) deduced from MS spectra (figure 5). Thus, the Fe–Cu bonds are larger than Fe–Fe and/or Fe–Ni bonds. That probably affects the peak position of Y (Fe1−x Cux )2 Si2 . Indeed, figure 14 shows that for x = 0.1, the M(T ) measured at 250 Oe, increases up to 300 K but no definite peak is observed. At 5 kOe, a shallow peak is obtained at 302(2) K. In addition, a sharp peak is observed at 24 K (not shown) its nature is not yet known. As expected, increase of Cu concentration shifts the peak position to lower temperatures, as shown in figure 15 for YFeCuSi2 (x = 0.5). The peak at 182 K (at H = 30 Oe) shifts to 170 and to 166 K for H = 250 Oe and 1 kOe respectively. 8

J. Phys.: Condens. Matter 26 (2014) 476002

I Felner et al

57

Fe M¨ossbauer studies of YFeMnSi2 at 90 K (main panel) and at 295 K (inset). Figure 18.

Figure 17. Temperature dependence of the magnetization of

YFeMnSi2 measured at 250 Oe. The inset shows three isothermal magnetization plots measured at 5, 90 and 290 K.

RT is also broad and is fitted with Heff = 39.2 kOe, indicating TM > RT, indicating the increase of Heff with x). Using the same token, we may analyze the magnetic curve for Y(Fe0.8 Mn0.2 )2 Si2 (figure 16) as composed of two magnetic contributions: the peak at 16 K arises from the Fe ions and the FM like behavior stems from the Mn ions. Due to the huge shift in the peak position caused by the magnetic Mn ions, we tend to believe that these two components coexist. If that is the case, the produced transferred hyperfine field is very small and is below the sensitivity of the M¨ossbauer technique. On the other hand, a phase separation case cannot be excluded. For this scenario, the absence of Heff on the Fe nuclei is obvious. We are turning now to the small deduced Heff = 30.3 kOe in YFeMnSi2 at 90 K. If the Fe had its own magnetic moment, then Heff should not be so sensitive to x and rather will be in full saturation with an order of magnitude higher Heff . But if the Heff is transferred from neighboring magnetic Mn ions, then each Fe which has different Mn ions as first neighbors will experience strongly reduced hyperfine fields. Assuming random distribution of Mn in the 3d layers, 94% of the Fe nuclei (given by (1 − (1 − x)4 )) should experience reduced fields which lead to the broadening of the spectrum at 90 K. Further calculations of Heff for other x values, is beyond the scope of this paper. However the Heff = 30.3 kOe obtained, fits well with the calculated Heff values for the Y (Fe1−x Mnx )2 Si2 system and to measured values on similar systems [32, 33, 35].

shift in the peak location is ∼160 K as compared to ∼40 K in Y(Fe0.85 Ni0.15 )Si2 . Further increase of x to 0.2, shifts the peak to 16 K at 250 Oe (figure 16) and to 13 K at 1 kOe. No peak is observed for the x = 0.3 sample. Figure 16 also shows for x = 0.2, a decrease in the M(T ) curve around 160 K which reminiscences a FM like behavior with a transition (TM ) ∼ 220 K. 57 Fe MS spectra for Y(Fe0.8 Mn0.2 )2 Si2 taken at 92 and 295 K (not shown) are identical, indicating an absence of any local magnetic field on Fe nuclei down to 92 K. Different features are observed in YFeMnSi2 . Figure 17 shows a sharp decrease in M(T ) curve up to 29 K and then a rise up to ∼100 K and a FM like behavior with TM ∼ 250 K. The increase of TM with x fits well with the magnetic phase transition obtained by ac susceptibility of Y (Mn1−x Fex )2 Si2 [33]. 57 Fe MS spectra taken at 90 K and RT indicate clearly, that the 90 K spectrum is much broader than the RT one (figure 18). Least-squares fits of the two spectra yield the hyperfine parameters as follows: the line width in both spectra is 0.33(3) mm s−1 . At RT the small quadrupole splitting = 0.12 ± 0.02 mm s−1 and the IS = 0.205 ± 0.002 mm s−1 obtained, agree well with those obtained for YFe2 Si2 presented above. The broader line at 90 K (IS = 0.30 mm s−1 ) was fitted by adding a magnetic hyperfine field parameter Heff = 30.3(3) kOe, a value which is an order of magnitude small than Heff = 330 kOe for pure Fe. That indicates that at 90 K, Fe in YFeMnSi2 senses transferred magnetic fields produced by the magnetic Mn ions. The sharp increase of the bulk magnetization below 29 K, is probably attributed to reorientation of the Mn magnetic sublattice a phenomenon which is very common to the RMn2 Si2 system [34]. A supporting evidence for this reorientation is given by the M(H ) plots shown in figure 17 (inset). Above the magnetic transition (at 290 K) a linear M(H ) curve is obtained. On the other hand, the two nonlinear M(H ) plots at 5 and 90 K, overlap at H > 25 kOe, but exhibit a quite different shape at low H values. At 5 K only, a small hysteresis loop is developed with a coercive field of 235 Oe. (For Y(Fe0.3 Mn0.7 )2 Si2 (x = 0.7) the 57 Fe MS spectrum at

4. Discussion and conclusions

Single or nearly single-phase polycrystalline samples of R (Fe1−x Nix )2 Si2 (R = La, Y and Lu), Y (Fe1−x Cux )2 Si2 and Y (Fe1−x Mnx )2 Si2 were synthesized and their magnetic properties have been measured. XRD patterns showed that all samples crystallize in the bct ThCr2 Si2 type structure. Because of the smaller ionic radius of Ni, the unit cell volume in all R (Fe1−x Nix )2 Si2 systems, decrease with x. On the other hand, for Y (Fe1−x Mnx )2 Si2 and Y (Fe1−x Cux )2 Si2 the unit cell volume increases with x. This is caused by the larger Mn radius and by the formal monovalent state of Cu in intermetallic compounds. 9

J. Phys.: Condens. Matter 26 (2014) 476002

I Felner et al

In both HTSC cuprates and Fe–As based materials the SC phase evolves from an AFM (or SDW) parent compound and it was proposed that the magnetic state is essential to mediate SC [4, 36]. As stated above, the similarity between the two BaFe2 As2 and RFe2 Si2 systems motivated us to look for SC at elevated temperature in various doped RFe2 Si2 materials, although Fe is not magnetic. The recent observation of bulk SC in YFe2 Ge2 at TC ∼ 1.8 K in [13], strengthens this intuition. None of the R (Fe1−x Mx )2 Si2 (M = Cu and Ni) samples listed above is SC above 1.8 K. This piece of evidence may shed some light on the SC mechanism, by proving the necessity of magnetic interactions in the parent compounds in order to induce SC. However, this assumption is not conclusive, because the full phase diagram of the R–Fe–Si is yet not known. One may argue that magnetic ordering might exist in compositions other than 1 : 2 : 2, therefore the absence of magnetic ordering in the stoichiometric 1 : 2 : 2 materials is just accidental. Two characteristic features in the dc magnetization studies are observed. (a) Pronounced peaks in the M(T ) plots for all systems (except for R = La). Four independent factors affect the peak position from which the first and the last factors are more drastic than the rest. (i) The peak is very sensitive to the lattice parameters. In RFe2 Si2, for R = La the peak is well above RT, but for R = Y and Lu (all measured at 250 Oe) the peak is at 226 K and 70 K respectively. In the doped R (Fe1−x Mx )2 Si2 samples, either (ii) increasing of x, (iii) applying magnetic fields or (iv) magnetic ions (such as Mn) all shift the peak to lower temperatures. The peak at 226 K for YFe2 Si2 is lowered by ∼40 in Y(Fe0.85 Ni0.15 )2 Si2 but ∼160 K in Y(Fe0.9 Mn0.1 )2 Si2 , and not observed down to 5 K in Y(Fe0.7 Mn0.3 )2 Si2 . No peak is observed in YNi2 Si2 its magnetic features are first reported here. Therefore, we attribute all peaks observed to the Fe sublattice in this structure. The peak position depends strongly of the Fe content. The absence of magnetic hyperfine field on the Fe nuclei (figure 5) in YFe2 Si2 (and in other compounds) confirms that Fe has no sizable long-range magnetic moment by its own. In Y (Fe1−x Mnx )2 Si2 the small transferred hyperfine field observed on Fe, is induced by the magnetic Mn sublattice. (b) A tiny extra FM phase with TC higher than RT, is obtained in all Y (Fe1−x Nix )2 Si2 samples whereas in the rest: Lu (Fe1−x Nix )2 Si2 , Y (Fe1−x Mx )2 Si2 (M = Cu or Mn) materials the impurity phase is negligible (figures 7, 14 and 15 and table 1). The absence of FM impurities in these systems excludes the possibility that the peaks observed are due to FM extra phase. The impurity phase probably due to tiny amount of unreacted Fe (or from YFe2 , YFe3 or Fex Siy ), contributes a constant moment to the various M(T ) plots and change their absolute values, but does not affect their features. Thus the main issue remains is the nature of the peaks observed in the systems studied here, which presumably all have the same origin. For the sake of simplicity we’ll discuss the magnetic behavior of YFe2 Si2 . The shortest Fe–Fe distance (given by 2−1/2 a) in YFe2 Si2 , is 2.77 Å. This distance is much longer than the critical value of 2.52 Å for FM interactions in Fe-based intermetallic compounds [37]. Thus no long-range magnetic order exists

in this sample at any temperature, as determined by our MS studies (figure 5). YFe2 Si2 was claimed to be AFM ordered at 275 K [19] or FM ordered at TC = 790 K [38]. Both determinations are shown to be wrong. The high reported TC just reflects the existing impurity phase (probably Fe3 Si). In previous publications, it was proposed that for non-magnetic R in RFe2 Si2, the divalent Fe ions are Pauli paramagnet with a virtually temperature-independent magnetic susceptibility [1, 39]. The present measurements using the high sensitive SQUID magnetometer reveals that Fe in YFe2 Si2 and in all its derivatives are rather PM. This is supported by (1) the high temperature M(T ) region which follows the CW law from which an average effective moment of Peff ∼ 3.1(1)µB /Fe is deduced. A similar Peff value was obtained in LuFe2 Ge2 [26]. (2) In all isothermal M(H ) curves the PM component is very pronounced. Peaks in the susceptibility around 50 and 400 K were observed in the iso-structural LuFe2 Ge2 [26] and LaNi2 Ge2 [27] but their origin remains unclear. A shallow peak is also observed in SrCu2 As2 but its appearance is ignored and not discussed [40]. Maximum in the susceptibility (around 70 K) was already observed for pure Pd metal which is considered as the classic example of NFFL, with large effect of spin fluctuations. With its high DOS at the Fermi energy and large Stoner exchange enhancement (∼10), Pd is easily polarized by dilute magnetic moments and often to FM at relatively high temperatures [41]. This maximum is a common feature in other NFFL compounds such YCo2 , and LuCo2 [42]. Quantitative calculation of the susceptibility is still a challenge even for the simplest case of Pd. YFe2 Zn20 (with Stoner enhancement factor of Z = 0.88) is closer to FM than Pd [43, 44] and also serves as an archetypical examples of NFFL. A faint maximum in the susceptibility around 10 K was obtained when measured at low magnetic fields. This maximum is monotonically decreased with increasing H . We tend to believe that YFe2 Si2 (similar to YFe2 Zn20 ) and its derivatives are additional examples for NFFL systems. In contrast to all examples mentioned, here the lattice parameters and/or the applied field, has a pronounced effect on the peak position. Doping of non-magnetic Ni ions also shift the peak to lower temperatures, but this shift is more dramatic when magnetic ions such as Mn are introduced. The current state of experiments does not allow us to suggest a consistent explanation to all phenomena presented here. We cannot exclude the possibility that nearly AFM or short-range AFM correlations are the origin of the peaks described here. Whatever the explanation is, the discovery of a pronounced peak in YFe2 Si2 is challenging for the theory of magnetism and propitious to further theoretical and experimental investigations. The study of these systems is a topic of ongoing research and promises to be a fruitful new phase space for several years to come. Acknowledgments

The research in Jerusalem was partially supported by the joint German–Israeli DIP project and in Houston by US AFOSR and TCSUH. We thank Professor I Nowik carrying out the M¨ossbauer studies. 10

J. Phys.: Condens. Matter 26 (2014) 476002

I Felner et al

[23] Turek I, Divis M, Niznansky D and Vejpravova J 2007 J. Magn. Magn. Mater. 316 403 [24] Lecocq P and Michel A 1964 Compt. Rend. 258 1817 [25] Mohn P and Schawarz K 1998 Physica B 130 26 [26] Ferstl J, Rosner H and Geibel C 2006 Physica B 378–380 744 Avila M A, Bud’ko S L and Canfield P C 2004 J. Magn. Magn. Mater. 270 51 [27] Goetsch R J, Anand V K, Pandey A and Johnston D C 2012 Phys. Rev. B 85 054517 [28] Braun H F 1980 Phys. Lett. A 75 386 [29] Kido H, Hoshikawa T, Shimada M and Koizumi M 1983 Phys. Status Solidi a 77 K121 [30] Sampathkumaran E V, Chaughule R S, Gopalaktishnan K V, Malik S K and Vijayaraghavan R 1983 J. Less Common Met. 92 35 [31] Nowik I, Levi Y, Felner I and Bauminger E R 1995 J. Magn. Magn. Mater. 147 373 [32] Meot-Meyer M, Venturini G, Malaman B, Mc Rae E and Rouques B 1985 Mater. Res. Bull. 20 1009 [33] Kervan S, Kilic A, Ozcan S and Gencer A 2004 Solid State Commun. 130 101 [34] Nowik I, Felner I and Bauminger E R 1997 Phys. Rev. B 55 3033 [35] Li H S, Cadogan J M, Zhao X I and Campbell S J J 1995 Magn. Magn. Mater. 147 91 [36] de la Cruz C et al 2008 Nature 453 899 [37] Friedel J, Lemand G and Olszewski S 1961 J. Appl. Phys. 32 3255 [38] Sankar S G, Malik S K, Rao V U S and Obermyer R 1976 AIP Conf. Proc. 34 236 [39] Gerasimov E G, Gaviko V S, Zaikov N K, Podgornykh S M, Korolev A V and Pirogov A N 2000 Phys. Met. Metallorg. 90 231 [40] Anand V K, Perera P K, Pandey A, Goetsch R J, Kreyssig A and Johnston D C 2012 Phys. Rev. B 85 214523 [41] Brodsky M B and Freeman A J 1980 Phys. Rev. Lett. 45 133 Gerhard W, Razavi F, Schilling J S, Huser D and Mydosh J A 1981 Phys. Rev. B 24 6744 [42] Burzo E, Gratz E and Pop V 1993 J. Magn. Magn. Mater. 123 159 [43] Jia S, Budko S L, Samolyuk G D and Canfield P C 2007 Nature Phys. 3 334 [44] Jia S Ni N, Bud’ko S L and Canfield P C 2009 Phys. Rev. B 80 104403

References [1] Felner I, Mayer I, Grill A and Schieber M 1975 Solid State Commun. 16 1005 [2] Pinto H and Shaked H 1973 Phys. Rev. B 7 3261 [3] Mihalik M, Svoboda P, Rusz J, Diviˇs M and Sechovsky V 2005 Physica B 367 19 [4] Kamihara Y, Watanabe T, Hirano M and Hosono H 2008 J. Am. Chem. Soc. 130 3296 [5] Sasmal K, Lv B, Lorenz B, Guloy A M, Chen F, Xue Y Y and Chu C W 2008 Phys. Rev. Lett. 101 107007 Rotter M, Tegel M and Johrendt D 2008 Phys. Rev. Lett. 101 107006 [6] Ni N, Bud’ko S L, Kreyssig A, Nandi S, Rustan G E, Goldman A I, Gupta S, Corbett J D, Kracher A and Canfield P C 2008 Phys. Rev. B 78 014507 [7] Sefat A S, Jin R, McGuire M A, Sales B C, Singh D J and Mandrus D 2008 Phys. Rev. Lett. 101 117004 [8] Alireza P L, Ko Y T C, Gillett J, Petrone C M, Cole J M, Lonzarich G G and Sebastian S E 2008 J. Phys.: Condens. Matter 21 012208 [9] Mani A, Ghosh N, Paulraj S, Bharathi A and Sundar C S 2009 Europhys. Lett. 87 17004 [10] Huang Q, Qiu Y, Ba W, Green M A, Lynn J W, Gasparovic Y C, Wu T, Wu G and Chen X H 2008 Phys. Rev. Lett. 101 257003 [11] Nandi S et al 2010 Phys. Rev. Lett. 104 57006 [12] Gooch M, Lv B, Lorenz B, Guloy A M and Chu C W 2009 Phys. Rev. B 79 104504 [13] Zou Y, Feng Z, Logg P W, Chen J, Lampronti G I and Grosche F M 2013 arXiv:1311.0247 [14] Hase I and Yanagisawa T 2011 Physica C 471 656 [15] Felner I and Nowik I 1978 J. Phys. Chem. Solids 39 767 Barandiaran J M et al 1987 J. Magn. Magn. Mater. 69 61 [16] Goto R, Noguchi S and Ishida T 2010 Physica C 470 S404 [17] Mine T, Yanagi H, Kamiya T, Kamihara Y, Hirano M and Hosono H 2008 Solid State Commun. 147 111 [18] Jeitschko W and Reehuis M 1987 J. Phys. Chem. Solids 48 667 [19] Shchebra I D, Antonov V M and Kotur B Ya 1996 J. Alloys Compounds 242 58 [20] www.periodictable.com/Properties/A/AtomicRadius.v.wt.html [21] Riger W and Parthe E 1969 Monatsh. Chem. 110 439 Iandelli A 1983 J. Less Common Met. 90 121 [22] Bauminger E R, Felner I, Froindlich D, Grill A, Lebenbaum D, Mayer I, Nowik I, Ofer S and Schieber M 1973 Tr. Mezhdunar, Konf. Nauka USSR 5 456

11